On Mon, 2007-10-01 at 06:41 -0400, Robert P. J. Day wrote:
<snip>
(Code snippets in my email are based on 2.6.22.6)
> so ELEMCOUNT() is clearly just an alternative to the more common
> ARRAY_SIZE() macro. but calling the macro SCTP_CHUNKMAP_IS_CLEAR()
> pretty clearly assumes that you can determine the array size of the
> macro parameter "chunkmap". however, from the source file
> net/netfilter/xt_sctp.c, we have:
> ...
> static inline bool
> match_packet(const struct sk_buff *skb,
> unsigned int offset,
> const u_int32_t *chunkmap, <----- only a *pointer*
> ...
This is called from one place - net/netfilter/xt_sctp.c in match():
...
&& SCCHECK(match_packet(skb, protoff + sizeof (sctp_sctphdr_t),
info->chunkmap, info->chunk_match_type,
info->flag_info, info->flag_count,
hotdrop),
...
At the start of the function, we have:
const struct xt_sctp_info *info = matchinfo;
In that struct, chunkmap is declared as:
u_int32_t chunkmap[256 / sizeof (u_int32_t)];
> switch (chunk_match_type) {
> case SCTP_CHUNK_MATCH_ALL:
> return SCTP_CHUNKMAP_IS_CLEAR(chunkmap);
> ...
>
> i don't see how you can reasonably call the macro
> SCTP_CHUNKMAP_IS_CLEAR() with a *pointer* to an array and expect that
> macro to be able to determine the "size" of that array. AFAIK,
> wouldn't you just get the answer "1" as the alleged size?
So while it's a pointer there, it was actually declared as an array.
Personally I don't like this kind of thing. I'd rather have an
array_size variable in the struct, but maybe that's just me.
> or is it a bad idea that i've been up all night again?
Get some sleep :-P
> rday
Avishay
--
To unsubscribe from this list: send an email with
"unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
Please read the FAQ at http://kernelnewbies.org/FAQ
